Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    3. Llama3.1-8B-IT (Instruct)
    4. Resid Post - 131k
    5. 11-RESID-POST-AA
    6. 85629
    Prev
    Next
    INDEX
    Explanations

    Questions and answers

    np_max-act · gemini-2.0-flash

    This neuron detects the imperative instruction “Read” (and its conjunction “and”) that begins user prompts asking the model to read and process text.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1
    Dataset (Dashboard)
    Various
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.11.hook_resid_post
    Architecture
    standard
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    _SUP
    -0.07
     Unlock
    -0.06
     FDA
    -0.06
     Hust
    -0.06
    _extensions
    -0.06
     Steps
    -0.06
     Draw
    -0.06
    ']}</
    -0.06
    	Set
    -0.06
    umpt
    -0.06
    POSITIVE LOGITS
     getName
    0.08
    ERENCE
    0.07
     unfavor
    0.06
    estruct
    0.06
    [
    0.06
    -clean
    0.06
    respect
    0.06
     Polymer
    0.06
    StartElement
    0.06
    ancia
    0.06
    Activations Density 0.018%

    No Known Activations